Advanced Leveraging of Government Data on Firms’ Vulnerability to Crisis Using Artificial Intelligence

Abstract

Government agencies possess large quantities of valuable data, which have to be leveraged to the highest possible extent, using the most advanced processing methods, in order to extract as much as possible the valuable knowledge they contain, in order to support future decisions, policies and programs. In this paper we develop a methodology for sophisticated advanced leveraging of data that government agencies possess concerning firms’ behaviour during recessionary economic crises as well as firms’ vulnerability to them, which is based on a combination of artificial intelligence (AI) techniques, both unsupervised and supervised ones. Economic crises are one of the most important significant challenges that market-based economies face, which has detrimental effects on businesses of most sectors. Governments employ diverse interventions, including extensive programs in order to mitigate the adverse effects of economic crises, which may include employment reduction, famine, public disturbance, and political instability. These government interventions, particularly the extensive economic stimulus programs during crises, can be rendered more efficacious by concentrating on the firms that are more vulnerable to economic crisis. In this direction the proposed methodology includes initially the use of unsupervised learning AI techniques (clustering) in order to identify based on the abovementioned government data the main typologies of firms with respect to the impacts of economic crisis they experience. Then it includes supervised learning AI techniques (classification) in order to predict based on these data the susceptibility/vulnerability of individual firms to future economic crises. Additionally, the authors present an initial implementation and substantiation (application) of the proposed methodology utilizing a dataset from the Greek Statistical Authority made available upon request, pertaining to 363 companies, data acquired during the Greek economic crises that occurred from 2009 to 2014. Satisfactory results were obtained from this first application.